Scientometrics
Ebrahim Ezzati Larsari; Ali Akbar Khasseh
Abstract
Background and Objectives: Among the features that have doubled the importance of the field of machine translation is its interdisciplinary nature; so that the influence of this field from sciences such as mathematics, statistics, probability, natural language processing, formal syntax, corpus linguistics ...
Read More
Background and Objectives: Among the features that have doubled the importance of the field of machine translation is its interdisciplinary nature; so that the influence of this field from sciences such as mathematics, statistics, probability, natural language processing, formal syntax, corpus linguistics and information technology is inevitable. Each of the aforementioned sciences has played a significant role in the emergence, formation and development of machine translation in some way. For this reason, researchers from different disciplines, each with their own expertise and interest, have turned to research in the field of machine translation and have caused the rapid development of this field. This study, employing a scientometric technique called Referenced Publication Years Spectroscopy (RPYS), concerns analyzing the most important historic works published in the area of Machine Translation.
Methodology: This study is conducted by a scientometric approach. Preliminary data of this study have been extracted from Web of Science. For this purpose, the references used in all the papers in Machine Translation since 1945 to the end of 2021 were studied. This search strategy led to 7899 records, covering 193912 references. Using RPYS software named CRExplorer, the revised data were analyzed.
Findings: Results exposed that the supreme enterprise of MT was initiated by Shannon in 1948; the one who illuminated the notion of entropy in information technology. After that, Weaver contributed to the appearance of statistical word by word MT in 1949. Moreover, it was Levenshtein (1966) who introduced the revolutionary notion of binary codes capable of correcting deletions, insertions, and reversals in MT. In addition, Dempster et al. (1977) presented a novel statistical EM algorithm concerning maximum likelihood from incomplete data. Later, the works of scholars Brown et al. (1993), Hochreiter (1997), Papineni et al. (2002), and Koehn et al. (2007) have been instrumental in shaping and developing MT.
Discussion: Based on the study of influential works in the field of machine translation, the role of mathematics and statistics in the formation of this field is evident. However, after advances in information technology and body linguistics, machine translation translates more accurately and conformingly to the source language.
